ES2396927T3 - Apparatus and procedure for decoding an encoded audio signal - Google Patents

Abstract

An apparatus for decoding (100) an encoded audio signal (102), the encoded audio signal (102) comprises a first portion (104a) encoded according to a first coding algorithm, a second portion (104b) encoded according with a second coding algorithm, BWE bandwidth extension parameters (106) for the first portion (104a) and the second portion (104b) and an encoding mode information (108) indicating a first decoding algorithm or a second decoding algorithm, comprising: A first decoder (110a) to decode the first portion (104a) according to the first decoding algorithm for a first time portion of the encoded signal (102) to obtain a first decoded signal (114a), wherein the first decoder (110a) comprises an LPC based encoder; A second decoder (110b) for decoding the second portion (104b) according to the second decoding algorithm for a second time portion of the encoded signal (102) to obtain a second decoded signal (114b), in which the second decoder (110b) comprises a transform-based encoder; A controllable BWE module (130) with crossover frequency (fx), the BWE module (130) configured to develop a bandwidth extension algorithm using the first decoded signal (114a) and the BWE parameters (106) for the first portion (104a), and to develop a 20 bandwidth extension algorithm using the second decoded signal (114b) and the bandwidth extension parameter (106) for the second portion (104b), in which the module BWE (130) is configured to use a crossover frequency for bandwidth extension for the first encoded signal (114a) and to use a second crossover frequency for bandwidth extension for the second decoded signal (114b) , 25 in which the first crossover frequency is greater than the second crossover frequency; and a controller (140) for controlling the crossover frequency (fx) for the BWE module (130) in accordance with the coding mode information (108).

Description

Apparatus and procedure for decoding an encoded audio signal

[0001] The present invention relates to an apparatus and a method for decoding an encoded audio signal, an apparatus for encoding, a method for encoding and an audio signal.

[0002] In the art, frequency domain coding schemes such as MP3 or AAC are known (English acronym for advanced audio coding translation). These frequency domain encoders are based on a time domain / frequency domain conversion, a subsequent quantization stage, where the quantization error is controlled using information from a psychoacoustic module, and an encoding stage, where the spectral coefficients quantized and the corresponding lateral information are encoded by entropy using code tables.

[0003] On the other hand, there are very appropriate encoders for voice processing such as AMR-WB + (English acronym for adaptive multiple speed translation - broadband) as described in 3GPP TS 26.290. Said speech coding schemes perform a Linear Predictive filter of a time domain signal. Said PL filter is derived from a Linear Predictive analysis of the time domain input signal. The resulting Linear Predictive filter coefficients are then quantized / encoded and transmitted as lateral information. The process is known as Linear Prediction Coding (LPC). At the output of the filter, the residual prediction signal or prediction error signal also known as the excitation signal is encoded using the analysis-by-synthesis stages of the ACELP encoder (acronym in English corresponding to the linear prediction translation with excitation by algebraic code) or, alternatively, it is encoded using a transformation encoder, which uses a Fourier transform with overlap. The decision between the ACELP coding (in English corresponding to the linear prediction translation with excitation by algebraic code) and the encoded excitation encoding of Transformation also called TCX coding (by its acronym in English) is made using a closed loop algorithm or open loop

[0004] Frequency domain audio coding schemes such as the AAC high-efficiency coding scheme (an acronym in English that corresponds to advanced audio encoder translation), which combines an AAC coding scheme and a replication technique Spectral band can also be combined with a joint or multi-channel stereo coding tool known under the name “MPEG surround” (acronym in English that corresponds to the translation Group of Experts in Motion Pictures

[0005] Said spectral band replication (SBR) comprises a technique that gained popularity as a complement to popular perception audio coding such as MP3 and advanced audio coding (AAC). . SBR comprises a procedure for bandwidth extension (BWE) where the low band (base band or core band) of the spectrum is encoded using an existing encoding, while the upper band (or high band) It is grossly parameterized using few parameters. SBR makes use of a correlation between the low band and the high band in order to predict the high band signal by extracting features from the lower band.

[0006] SBR is, for example, used in HE-AAC or AAC + SBR. In SBR it is possible to dynamically change the crossover frequency (BWE start frequency) as well as the temporal resolution implied by the number of parameter sets (envelope) per frame. AMR-WB + implements a time domain bandwidth extension combined with a time / frequency switched core core decoder, providing good audio quality especially for voice signals. A limiting factor for AMR-WB + audio quality is the audio bandwidth common to both BWE core encoders and start-up encoders that represents a quarter of the system's internal sampling rate. While the ACELP voice model is capable of modeling voice signals well enough throughout the bandwidth, the frequency domain audio encoder fails to send decent quality for some general audio signals. Consequently, voice coding schemes show high quality for voice signals even at low bit rates, but show poor quality for music signals at low bit rates.

[0007] Domain frequency coding schemes such as HE-AAC are advantageous because they show high quality at a low bit rate for music signals. However, the quality of voice signals at a low bit rate is problematic.

[0008] Therefore, different kinds of audio signal demand different features of the bandwidth extension tool. WO 02/41302 A and 2008/031458 A1 describe approaches to bandwidth extension.

[0009] The objective of the present invention is to provide an improved coding / decoding concept.

[0010] The objective is achieved by means of an audio decoder according to claim 1, a method for decoding audio according to claim 13, an encoder according to claim 8, a method for encoding according to the claim 14, an encoded signal according to claim 15 or a computer program according to claim 16.

[0011] The present invention is based on discovering that the crossover frequency or the BWE start frequency is a parameter that influences the audio quality. While time domain (voice) encoders usually encode the entire frequency range for a given sampling rate, the audio bandwidth is a tuning parameter (for example music encoders), which decreases the total amount of spectral lines to encode and at the same time the amount of bits per spectral line available to encode will increase, meaning that an exchange of quality versus audio bandwidth is performed. Therefore, in the new approach, different core encoders are combined with variable audio bandwidths with a system switched with a common BWE module, where the BWE module must represent the different audio bandwidths.

[0012] A direct way would be to find the smallest of all the bandwidths of the core encoders and use it as the BWE start frequency, but this would deteriorate the perceived audio quality. In addition, the encoding efficiency would be reduced, since in time sections where a core encoder is active which has a greater bandwidth than the BWE start frequency, some frequency regions would be represented twice, by the encoder of core itself as BWE that introduces redundancy. A better solution is therefore to adapt the BWE start frequency to the audio bandwidth of the core encoder used.

[0013] Therefore, according to the embodiments of the present invention an audio coding system combines a bandwidth extension tool with a core encoder that depends on the signal (eg voice encoder / switched audio), where the crossover frequency comprises a variable parameter. A signal sorting output that controls the switching between different core coding modes can also be used to switch the characteristics of the BWE system such as temporal resolution and blur, spectral resolution and crossover frequency.

[0014] Therefore, one aspect of the present invention consists of an audio decoder for an encoded audio signal, the encoded audio signal comprises a first portion encoded according to a first coding algorithm, a second portion encoded according to with a second encoding algorithm, BWE parameters for the first portion and the second portion and encoding mode information indicating a first decoding algorithm or a second decoding algorithm, comprising a first decoder, a second decoder, a BWE module and a controller The first decoder decodes the first portion according to the first decoding algorithm for a first time portion of the encoded signal to obtain a first decoded signal. The second decoder decodes the second portion according to the second decoding algorithm for a second time portion of the encoded signal to obtain a second decoded signal. The BWE module has a controllable crossover frequency and is configured to perform a bandwidth extension algorithm using the first decoded signal and BWE parameters for the first portion, and to perform a bandwidth extension algorithm using a second decoding signal and the bandwidth extension parameter for the second portion. The controller controls the crossover frequency for the BWE module according to the encoding mode information.

[0015] According to another example, an apparatus for encoding an audio signal comprising a first and a second encoder, a decision stage and a BWE module. The first encoder is configured to encode according to a first encoding algorithm the first encoding algorithm with a first frequency bandwidth. The second encoder is configured to encode according to a second encoding algorithm, the second encoding algorithm has a second frequency bandwidth less than the first frequency bandwidth. The decision stage indicates the first coding algorithm for a first portion of the audio signal and the second coding algorithm for a second portion of the audio signal, the second portion is different from the first portion. The bandwidth extension module calculates the BWE parameters for the audio signal, where the BWE module is configured to be controlled by the decision stage to calculate the BWE parameters for a band that does not include the first frequency bandwidth in the first portion of the audio signal and for a band that does not include the second frequency bandwidth in the second portion of the audio signal.

[0016] In contrast to the embodiments, SBR in prior art is applied to an audio encoder without a switch that only results in the following disadvantages. The temporal resolution as well as the crossover frequency could be applied dynamically, but cutting-edge implementations such as the 3GPP source apply only a temporary resolution change for bit rates such as castanets. Likewise, a finer total temporal resolution could be chosen at higher speeds as a tuning parameter that depends on or a decision threshold value that controls the temporal resolution, which best combines the bit rate of the signal. There is no explicit classification that determines the type of temporal resolution, such as stationary tonal music versus voice. The embodiments of the present invention overcome these disadvantages. The embodiments especially allow an adapted crossover frequency combined with a flexible option for the core encoder used for the encoded signal to provide significantly higher quality compared to a prior art encoder / decoder.

Brief description of the drawings

[0017] Preferred embodiments of the present invention are described below with respect to the accompanying drawings, where:

Fig. 1 shows a block diagram of an apparatus for decoding;

Fig. 2 shows a block diagram of an apparatus for decoding according to the first aspect of the present invention;

Fig. 3 shows a block diagram of a coding scheme in greater detail;

Fig. 4 shows a block diagram of a decoding scheme in greater detail;

Fig. 5 shows a block diagram of an encoding scheme according to a second aspect;

Fig. 6 is a schematic diagram of a decoding scheme according to the second aspect;

Fig. 7 illustrates an LPC stage of the encoder side that provides short-term prediction information and the prediction error signal;

Fig. 8 illustrates another embodiment of an LPC device for generating a weighted signal;

Figs. 9a-9b show an encoder comprising an audio / voice switch that results in different temporal resolutions for an audio signal;

Fig. 10 illustrates a representation for an encoded audio signal.

Detailed description of the invention

[0018] Fig. 1 shows a decoder apparatus 100 for decoding an encoded audio signal 102. The encoded audio signal 102 comprises a first portion 104a encoded according to the first encoding algorithm, a second portion 104b encoded in accordance with a second encoding algorithm, a BWE parameter 106 for the first time portion 104a and the second time portion 104b and an encoding mode information 108 indicating a first decoding algorithm or a second decoding algorithm for the respective portions of weather. The decoding apparatus 100 comprises a first decoder 110a, a second decoder 110b, a BWE module 130 and a controller

140. The first decoder 110a is adapted to decode the first portion 104a according to the first decoding algorithm for the first time portion of the encoded signal 102 to obtain a first decoded signal 114a. The second decoder 110b is configured to decode the second portion 104b according to the second decoding algorithm for a second time portion of the encoded signal to obtain a second decoded signal 114b. The BWE 130 module has a controllable crossover frequency fx that adjusts the behavior of the BWE 130 module. The BWE 130 module is configured to develop a bandwidth extension algorithm to generate audio signal components in the upper frequency band based on the first decoding signal 114a and the BWE parameters 106 for the first portion, and for generating components of the audio signal in the upper frequency band based on the second decoded signal 114b and the bandwidth extension parameter 106 For the second portion. The controller 140 is configured to control the crossover frequency fx of the BWE module 130 in accordance with the information of the coding mode 108.

[0019] The BWE module 130 may also comprise a combiner that combines the components of the audio signal of the lower and upper frequency band and emissions of the resulting audio signal 105.

[0020] The information of the encoding mode 108 indicates, for example, which portion of time of the encoded audio signal 102 is encoded by which encoding algorithm. This information can at the same time identify the decoder to be used for the different portions of time. In addition, the information of the coding mode 108 can control a switch to switch between different decoders for different time portions.

[0021] Therefore, the crossover frequency fx is an adjustable parameter that is adjusted according to the decoder used which can, for example, comprise a voice encoder as the first decoder 110a and an audio decoder as the second decoder 110b. As stated above, the crossover frequency fx for a voice decoder (such as LPC based) may be greater than the crossover frequency used for an audio decoder (for example for music). Accordingly, in other embodiments the controller 220 is configured to increase the crossover frequency fx or to decrease the crossover frequency fx within one of the time portions (for example the second time portion) so that the frequency of Crossover can be changed without changing the decoding algorithm. This means that a change in the crossover frequency may not be related to a change in the decoder used: the crossover frequency can be changed without changing the decoder used or vice versa the decoder can be changed without changing the crossover frequency.

[0022] The BWE module 130 may also comprise a switch controlled by the controller 140 and / or by the BWE parameter 106 so that the first decoded signal 114a is processed by the BWE module 130 during the first portion of time and the second signal of Decoding 114b is processed by the BWE 130 module during the second portion of time. This switch can be activated by a change in the crossover frequency fx or by an explicit bit within the encoded audio signal 102 indicating the coding algorithm used during the respective portion of time.

[0023] In other embodiments, the switch is configured to switch between the first and second portion of time from the first decoder to the second decoder so that the bandwidth extension algorithm is applied to the first decoded signal or the second signal. decoded. Alternatively, the bandwidth extension algorithm is applied to the first and / or second decoded signal and the switch locates thereafter so that one of the extended bandwidth signals falls.

[0024] Fig. 2 shows a block diagram for an apparatus 200 encoding an audio signal 105. The encoder apparatus 200 comprises a first encoder 210a, a second encoder 210b, a decision stage 220 and an expansion module of bandwidth (BWE module) 230. The first encoder 210a is operative to encode according to a first encoding algorithm with a first frequency bandwidth. The second encoder 210b is operative to encode according to a second encoding algorithm with a second frequency bandwidth less than the first frequency bandwidth. The first encoder may, for example, be a voice encoder like the LPC-based encoder, while the second encoder 210b may comprise an audio (music) encoder. Decision stage 220 is configured to indicate the first coding algorithm for the first portion 204a of the audio signal 105 and to indicate the second coding algorithm for a second portion 204b of the audio signal 105, where the second portion of Time is different from the first portion of time. The first portion 204a may correspond to the first portion of time and the second portion 204b may correspond to the second portion of time which is different from the first portion of time.

[0025] The BWE 230 module is configured to calculate the BWE 106 parameters for the audio signal 105 and is configured to be controlled by the decision stage 220 to calculate the BWE 106 parameter for a first band that does not include the first width of frequency band in the first time portion 204a of the audio signal 105. The BWE module 230 is further configured to calculate the BWE parameter 106 for a second band that does not include the second bandwidth in the second time portion 204b of the audio signal 105. The first (second) band therefore comprises frequency components of the audio signal 105 that are outside the first (second) frequency bandwidth and are limited to the lower end of the spectrum by the frequency of crossing fx. The first or second bandwidth can therefore be defined by a variable crossover frequency controlled by decision step 220.

[0026] In addition, the BWE module 230 may comprise a switch controlled by the decision stage 220. The decision stage 220 may determine a preferred coding algorithm for a given portion of time and controls the switch so that during the portion of time given use the preferred encoder. The

Modified coding mode information 108 'comprises the corresponding switching signal.

In addition, the BWE module 230 may also comprise a filter to obtain components of the audio signal 105 in the lower / upper frequency band that are separated by the crossover frequency fx which may comprise a value of approximately 4 kHz or 5 kHz. Finally, the BWE module 130 may also comprise an analysis tool for determining the BWE parameter 106. The information on the mode of

Modified coding 108 ’can be equivalent (or equal) to the information in coding mode 108. The

encoding mode information 108 indicates, for example, the encoding algorithm used for the respective portions of time in the bit sequence of the encoded audio signal 105.

[0027] According to other examples, decision step 220 comprises a signal classification tool that analyzes the original input signal 105 and generates the control information 108 that causes the selection of the different coding modes. The analysis of the input signal 105 depends on the objective of choosing the optimal core coding mode for a given input signal frame. The output of the signal classifier can (optionally) also be used to influence the behavior of other tools, for example, sonic surround MPEG, extended SBR, filter bank with time interleaving and others. The input to the signal sorting tool comprises, for example, the original unmodified input signal 105, but also parameters that depend on the additional implementation optionally. The output of the signal sorting tool comprises the control signal 108 to control the selection of the core encoder (for example unfiltered frequency domain LP or LP filtered time or frequency domain coding or other coding algorithms).

[0028] According to the examples, the crossover frequency fx is a dependent set signal combined with the switching decision to use a different coding algorithm. Therefore, a simple switching signal can simply be a change (a jump) in the crossover frequency fx. In addition, the information of the coding mode 108 may also comprise the change in crossover frequency fx which indicates at the same time a preferred coding scheme (eg voice / audio / music).

[0029] According to examples the decision step 220 is operative to analyze the audio signal 105 or a first output of the first encoder 210a or a second output of the second encoder 210b or a signal obtained by decoding an output signal of the encoder 210a or the second encoder 210b with respect to the destination function. The decision stage 220 may optionally be operative to perform a voice / music discrimination such that a voice decision is favored with respect to the music decision for the voice decision to be made, for example, even when a portion less than 50% of a frame for the first switch is voice and a portion greater than 50% of the frame for the first switch is music. Therefore, the decision stage 220 may comprise an analysis tool that analyzes the audio signal to decide whether the audio signal is primarily a voice signal or primarily a music signal so that taking into account the result of the stage The decision can decide which is the best encoder to use for the analyzed time portion of the audio signal.

[0030] Figs. 1 and 2 do not show many of these details for the encoder / decoder. Some possible detailed examples for the encoder / decoder are shown in the following figures. Furthermore, with respect to the first and second decoder 110a, b of Fig. 1 other decoders may be present which may or may not use for example other coding algorithms. Similarly, the encoder 200 of Fig. 2 can also comprise additional encoders that can use additional coding algorithms. In the following example with two encoders / decoders will be explained in greater detail.

[0031] Fig. 3 illustrates in greater detail an encoder with two cascade switches. A mono signal, a stereo signal or multi-channel signal is entered in decision stage 220 and in a switch 232 that is part of the BWE module 230 of Fig. 2. Switch 232 is controlled by decision stage 220. Alternatively, decision stage 220 may also receive lateral information included in the mono signal, the stereo signal or the multi-channel signal or is at least associated with said signal, where the information is existing, which was, by example, generated by originally producing the mono signal, the stereo signal or the multi-channel signal.

[0032] Decision step 220 activates the switch in order to deliver a signal in a frequency coding portion 210b illustrated in an upper branch of Fig. 3 or a domain coding portion LPC-210a illustrated in a branch lower in Fig. 3. A key element of the frequency domain coding branch consists of a spectral conversion block 410 which is operative to convert a common pre-processing stage output signal (as discussed below. ) in a spectral domain. The spectral conversion block may include an MDCT algorithm, a QMF algorithm, an FFT algorithm, a Wavelet analysis or a filter bank such as a filter bank taken by sampling critically with a certain number of filter bank channels, where Sub-band signals in this filter bank can be signals of real value or complex value. The output of the spectral conversion block 410 is encoded using a spectral audio encoder 421, which may include processing blocks as known from the AAC coding scheme.

[0033] Generally, processing at branch 210b is a processing in a model based on the perception or sink information model. In that way, this branching models the sink sound of the human auditory system. In contrast, processing at branch 210a must generate a signal in the excitation, residual or LPC domain. Generally, the processing at branch 210a is a processing of a voice model or information generating model. For voice signals, this model is a human voice model / sound generating system.

However, if a sound from a different source that requires a different sound generating model must be encoded, then the processing at branch 210a may be different.

In addition to the coding branches shown, other examples comprise additional branches.

or core encoders. For example, different encoders may optionally be present for different sources, so that the sound of each source can be encoded by using a preferred encoder.

[0034] In the lower coding branch 210a, a key element consists of an LPC 510 device, which emits an LPC information used to control the characteristics of an LPC filter. This LPC information is transmitted to a decoder. The output signal 510 of the LPC step is an LPC domain signal consisting of an excitation signal and / or a weighted signal.

[0035] The LPC device generally issues an LPC domain signal, which can be any signal in the LPC domain or other signal, which has been generated by applying LPC filter coefficients to an audio signal.

Also, an LPC device can also determine these coefficients and can also quantify / encode these coefficients.

[0036] The decision in the decision stage 220 may be an adaptive signal for the decision stage to develop a music / voice discrimination and control the switch 232 such that the musical signals are entered in the upper branch 210b, and the voice signals are entered in the lower branch 210a. In one example, decision stage 220 supplies its decision information in a sequence of output bits so that a decoder can use this decision information in order to carry out the correct decoding operations. This decision information may, for example, comprise the information of the coding mode 108 which may also comprise information on the crossover frequency fx or a change in the crossover frequency fx.

[0037] Said decoder is illustrated in Fig. Fig. 4. The emission of the signal by the spectral audio encoder 421 is, after transmission, input into a spectral audio decoder 431. The signal emission of the decoder of spectral audio 431 is entered into a time domain converter 440. (the time-domain converter may in general be a converter from the first to the second domain) Similarly, the issuance of the LPC 210a domain coding branch of Fig. 3 is received on the decoder side and processed by elements 531, 533, 534, and 532 to obtain an LPC excitation signal. The excitation of the LPC signal is entered in a synthesis stage LPC 540, which receives, as an additional input, the LPC information generated by the corresponding LPC analysis stage 510. The output of the time-domain converter 440 and / or the output of the synthesis stage LPC 540 are entered in a switch 132 which can be part of the BWE module 130 in Fig. 1. The switch 132 is controlled through a switch control signal (such as the information of the coding mode 108 and / or parameter BWE 106) that was, for example, generated by decision step 220, or that was provided externally by means of a creator of the mono signal, stereo signal or original multi-channel signal. Switch 600 signal is a mono signal, stereo signal or full multi-channel signal

[0038] In Fig. 3 The input signal in the switch 232 and the decision stage 220 may be a mono signal, stereo signal or multi-channel signal or generally an audio signal. Depending on the decision that may be derived from the input signal of switch 232 or any external source as a producer of the original audio signal underlying the signal emission in step 232, the switch switches between the frequency coding branch 210b and the LPC 210a coding branch. The frequency coding branch 210b comprises a spectral conversion stage 410 and a subsequent connected quantization / coding stage 421. The quantization / coding stage may include any of the known functionalities of modern frequency domain encoders such as the AAC encoder. . In addition, the quantization operation in the quantization / coding stage 421 can be controlled through a psychoacoustic module that generates psychoacoustic information such as the threshold value of psychoacoustic masking on frequency, where this information is entered in step 421.

[0039] In the LPC210a coding branch, the switch output signal is processed through the LPC 510 analysis stage that generates information from the LPC side and an LPC domain signal. The excitation encoder inventively comprises an additional switch for switching the further processing of the LPC domain signal between the quantization / coding operation 522 in the LPC domain or a quantization / coding step 524, which processes values in the LPC spectral domain. For this purpose, a spectral converter 523 is provided at the input of the quantization / coding step 524. The switch 521 is controlled by open loop mode or closed loop mode depending on the specific configurations, for example, are described in The technical specification AMR-WB +.

[0040] For the closed loop control mode, the encoder further includes a quantizer / reverse encoder 531 for the LPC domain signal, a quantizer / reverse encoder 533 for a LPC spectral domain signal and a reverse spectral converter 534 for the item 533 is output. Both encoded and newly decoded signals in the processing branches of the second coding branch are input into the switch control device 525. In the switch control device 525, these two output signals are compared between yes and / or with a destination function or a destination function is calculated which can be based on a comparison of the distortion in both signals so that the signal with less distortion is used to decide, what position the switch 521 should take. alternatively, if both branches provide non-constant bit rates, the branch that provides the bi rate The smaller ts could be selected even when the perceptual distortion or distortion of this branch is less than the perceptual distortion or distortion of the other branch (an example for the distortion may be a signal-to-noise ratio). Alternatively, the destination function could use, as input, the signal-to-noise ratio of each signal and a bit rate of each signal and / or additional criteria in order to find the best decision for a specific objective. If, for example, the objective is such that the bit rate should be as low as possible, then the target function would be based largely on the bit rate of the emission of the two signals by means of elements 531, 534. However, when the main objective is to obtain the best quality for a certain bit rate, then the control of switch 525 could, for example, discard each signal that is above the allowed bit rate when both signals meet. Below the allowed bit rate, the switch control would select the signal with the best signal to noise ratio, that is, with the least quantization / coding distortion.

[0041] The coding scheme according to the present invention, as previously established, is illustrated in Fig. 4. For each type of the three output signals, there is a specific decoding / re-quantization step 431, 531 or 533. While step 431 emits a time spectrum that becomes a time domain using the frequency / time converter 440, step 531 emits an LPC domain signal, and item 533 emits an LPC spectrum. In order to ensure that the input signals on the switch 532 are both in the LPC domain, the LPC spectrum / LPC converter 534 is provided. The output data of the switch 532 is again transformed into the time domain using a step of synthesis LPC 540, controlled through LPC information generated and transmitted from the encoder side. Then, subsequent to block 540, both branches have switched time domain information in accordance with a switching control signal in order to finally obtain an audio signal such as a mono signal, stereo signal, or multi-channel signal. , which depends on the signal input in the coding scheme of Fig. 3.

[0042] Figs. 5 and 6 show other examples for the encoder / decoder, where the BWE stages as part of the BWE modules 130, 230 represent a common processing unit.

[0043] Fig. 5 illustrates an encoding scheme, where the common preprocessing scheme connected to the input of the switch 232 may comprise a stereo surround / assembly block 101 which generates, as output, joint stereo parameters and a signal of Mono output, generated by mixing down the input signal that is one with two or more channels. Generally, the signal at the output of block 101 may also be a signal with more channels, but due to the downward mixing functionality of block 101, the amount of channels at the output of block 101 will be less than the number of channels in the block 101 entry.

[0044] The common preprocessing scheme may comprise in addition to block 101 a step of bandwidth extension 230. In the example of Fig. 5 the output of block 101 is input to the bandwidth block 230 which emits a limited band signal such as the low band signal or low pass signal at its output. Preferably, this signal is also subjected to subsampling (for example by a factor of two). Bandwidth extension parameters such as spectral envelope parameters, reverse filtering parameters, noise floor parameters etc. as they are known from the MPEG-4 HE-AAC profile they are generated and emitted to an 800 bit sequence multiplexer.

[0045] Preferably, the decision stage 220 receives the signal input in block 101 or input in block 230 in order to decide between, for example, a music mode or voice mode. In music mode, the upper coding branch 210b (second encoder in Fig. 2) 210a is selected while in voice mode, the lower coding branch 101 and / or the bandwidth extension block is selected 230 to adapt the functionality of these blocks to the specific signal. Thus, when the decision stage 220 determines that a certain portion of time of the input signal corresponds to the first mode as the music mode, the specific characteristics of block 101 and / or block 230 can be controlled by decision stage 220 Alternatively, when the decision stage 220 determines that the signal is in a voice mode or, generally, in a second LPC domain mode, the specific characteristics of the blocks 101 and 230 can be controlled according to the output of the decision stage. The decision step 220 also produces the control information 108 and / or the crossover frequency fx that can also be transmitted to the BWE block 230 and, in addition, to the bit sequence multiplexer 800 to be transmitted to the decoder side.

[0046] Preferably, the spectral conversion of the coding branch 210b is performed using an MDCT operation which, more preferably, consists of the time-distorted MDCT operation, where the force or, generally, the distortion force (warp ) can be controlled between zero and a high distortion force (warp). At a zero distortion force (warp), the MDCT operation in block 411 is a simple MDCT operation known in the art. The interleaving force in time together with the lateral interleaving information in time can be transmitted / entered into a bit sequence multiplexer 800 as lateral information.

[0047] In the LPC coding branch, the LPC domain encoder may include an ACELP 526 core that calculates a tone gain, a tone delay and / or information book information as an index of the information book and gain of the code The TCX mode as known from 3GPP TS 26.290 incurs a perceptually weighted signal processing in the transformation domain. A weighted Fourier transformed signal is quantized using a networked quantization with split multiple velocity (algebraic VQ) with noise factor quantization. A transformation is calculated at 1024, 512, or 256 sample windows. The excitation signal is recovered by inverse filtering of the quantized weighted signal through a weighting filter. The TCX mode can also be used in modified form in which MDCT is used with extended overlap, scalar quantization, and an arithmetic encoder to encode spectral lines.

[0048] In the "music" 210b coding branch, a spectral converter preferably comprises

an MDCT operation specifically adapted MDCT with certain window functions followed by a quantization / entropy coding step which may consist of a quantization step with a single vector, but preferably is a combined quantization / entropy scalar encoder similar to the encoder of

5 quantization / entropy in the frequency domain coding branch, that is, in item 421 of Fig. 5.

[0049] In the "voice" coding branch 210a, there is the LPC 510 block followed by a 521 switch, again followed by an ACELP 526 block or TCX 527 block. ACELP is described in 3GPP TS

26.190 and TCX is described in 3GPP TS 26.290. Generally, the ACELP 526 block receives an LPC excitation signal calculated by a procedure described in Fig. 7. The TCX 527 block receives a weighted signal generated by Fig. 8. [0050] On the decoder side illustrated in Fig. .6, after the inverse spectral transformation in the block 537, the inverse of the weighting filter is applied, that is (1--1) / (1-A (z /)). Then, the signal is filtered through (1-A (z)) to address the LPC excitation domain. Thus, the conversion to the 15 domain block LPC 534 and the TCX-1 537 block include inverse transformation and then filtering through

to convert from the weighted domain to the excitation domain.

[0051] Although item 510 in Figs. 3, 5 illustrates a single block, block 510 can emit different signals as long as these signals are in the LPC domain. The actual mode of block 510 such as the excitation signal mode or the weighted signal mode may depend on the actual state of the switch. Alternatively, block 510 may have two parallel processing devices, where one device is implemented in a manner similar to Fig. 7 and the other device is implemented as in Fig. 8. Therefore, the

The LPC domain at the output of 510 may represent the LPC excitation signal or the LPC weighted signal, or other LPC domain signal.

[0052] In the second coding branch (ACELP / TCX) of Fig. 5, the signal is preferably pre- emphasized through a filter 1--1 before coding. In the ACELP / TCX decoder in Fig. 6 the synthesized signal is de-emphasized with the filter 1 / (1-z-1). In a preferred example, the parameter m has the value 0.68. The pre-emphasis may be part of the LPC 510 block where the signal is pre-emphasized before the LPC analysis and quantization. Similarly, the de-emphasis can be part of the LPC-1 block synthesis block

540

[0053] Fig. 6 illustrates a decoding scheme corresponding to the coding scheme of Fig. 5. The bit sequence generated by a bit sequence multiplexer 800 (or output interface) of Fig. 5 is

35 entered in a 900 bit sequence multiplexer (or input interface). Depending on the information derived for example from a sequence of bits through a mode detection block 601 (for example part of the controller 140 in Fig. 1), it is controlled to emit signals from the upper branch or signals from the lower branch to the bandwidth extension block 701. The bandwidth extension block 701 receives, from the bit stream demultiplexer 900, side information and, based on this side information and the issuance of the mode decision 601, reconstructs the high band based on the low band output by the switch 600. The control signal 108 controls the crossover frequency fx used.

[0054] The full band signal generated by block 701 is input into the joint stereo / surround processing stage 702, which reconstructs two stereo channels or several multi-channels. Generally, block 702 will broadcast more channels that were entered into this block. Depending on the application, the

The entry in block 702 may still include two channels as in stereo mode and may even include more channels as long as the broadcast on this block has more channels than the entry in this block.

[0055] It has been observed that switch 232 in Fig. 5 switches between both branches so that only one branch receives a signal to process and the other branch does not receive a signal to process. In an example alternative, however, the switch 232 may also be arranged subsequently to, for example, the audio encoder 421 and the excitation encoder 522, 523, 524, which means that both branches 210a, 210b process the Same signal in parallel. In order not to double the bit rate, however, only the signal output by one of those encoding branches 210a or 210b is selected to be written in the sequence of output bits. The decision stage will operate so that the signal written in the bit sequence minimizes a certain cost function, where the cost function may be the generated bit sequence or the perceptual distortion generated a combined cost function of

speed / distortion Consequently, either in this mode or in the mode illustrated in the Figures, the decision stage can also operate in the closed loop mode in order to ensure that, finally, only the coding branching output is written in the bit sequence at which it has for a lower perceptual distortion given the lowest bit rate or, for a given bit rate, the lowest perceptual distortion. In closed loop mode, the feedback input can be derived from the emissions of the three quantizer / scaler blocks 421, 522 and 424 in Fig. 3.

[0056] Also in the embodiment of Fig. 6, the switch 132 may in alternative embodiments be arranged after the BWE module 701 so that the bandwidth extension is performed in parallel for both branches and the switch selects one of the two extended bandwidth signals.

[0057] In the implementation with two switches, that is, the first switch 232 and the second switch 521, it is preferable that the time resolution for the first switch is less than the time resolution for the second switch. Expressed differently, the blocks of the input signal in the first switch, which can be switched through a switch operation, are larger than the blocks switched by the second switch 521 operating in the LPC domain. For example, the frequency domain / LPC domain switch 232 can switch blocks of a length of 1024 samples, and the second switch 521 can switch blocks with 256 samples each.

[0058] Fig. 7e illustrates a more detailed implementation of the LPC 510 analysis block. The audio signal is input into a filter determination block that determines the information of the filter A (z). This information is issued as short-term prediction information necessary for a decoder. The short-term prediction information is required by the actual prediction filter 85. In a subtractor 86, a current sample of the audio signal is input and a predictable value for the current sample is subtracted so that for this sample, the signal Prediction error is generated on line 84.

[0059] While Fig. 7e illustrates a preferred way to calculate the excitation signal, Fig. 7f illustrates a Preferred way to calculate the weighted signal. In contrast to Fig. 7e, filter 85 is different, when it is different from 1. A value less than 1 is preferred for. Also, block 87 is present, and is preferably a number less than 1. Generally, the elements in Fig. 7e and 7f can be implemented as in 3GPP TS 26,190 or 3GPP TS 26,290.

[0060] Subsequently, a synthesis analysis CELP encoder will be analyzed in relation to Fig. 6 in order to illustrate the modifications applied to this algorithm. This CELP encoder is discussed in detail in “Voice Coding: A Tutorial Review,” Andreas Spanias, IEEE Procedures, Vol. 82, No. 10, October 1994, pages 1541-1582.

[0061] For specific cases, when a picture is a mixture of vocalized and non-vocalized speech or when voice over music occurs, a TCX encoding may be more appropriate for encoding the excitation in the LPC domain. The TCX encoding processes the weighted signal in the frequency domain without assuming excitation productions. TCX is more generic than CELP coding and is not restricted to a vocalized or non-vocalized excitation source model. TCX is still a coding with a filter-source model that uses a linear prediction filter to model the formants of voice type signals.

[0062] In the AMR-WB + - type coding, a selection is made between the different TCX and ACELP modes as known from the one described AMR-WB +. The TCX modes are different since the length of the Discrete Fourier Transform in the block direction is different for different modes and the best mode

It can be selected using a synthesis analysis approach or direct “pre-feeding” mode.

[0063] As discussed in relation to Figs. 5 and 6, the common preprocessing stage 100 preferably includes a joint multi-channel device (joint / stereo joint) 101 and, in addition, a step of widening the width of band 230. Correspondingly, the decoder includes a bandwidth extension stage 701 and a set multichannel stage subsequently connected 702. Preferably, the set multichannel stage 101 is, with respect to the encoder, connected before the extension stage of bandwidth 230, and, on the decoder side, the bandwidth extension stage 701 is connected before the set multichannel stage 702 with respect to the signal processing direction. Alternatively, however, the common preprocessing stage may include a joint multichannel stage without subsequent bandwidth extension or bandwidth extension stage without a connected multichannel stage connected.

[0064] Figs. 9a to 9b show a simplified view in the encoder of Fig. 5, where the encoder comprises the switching decision unit 220 and the stereo coding unit 101. In addition, the encoder also comprises bandwidth extension tools 230 such as , for example, an envelope data calculator and modules related to SBR. The switching decision unit 220

provides a switching decision signal 108 ’that switches between audio encoder 210b and the

210a voice encoder The voice encoder 210a can also be divided into a vocalized and a non-vocalized encoder. Each encoder can encode the audio signal in the core frequency band using different numbers of sample values (for example, 1024 for a higher resolution or 256 for a lower resolution). The switching decision signal 108 ’is also supplied to the bandwidth extension tool (BWE) 230. The BWE 230 tool will use the 108’ decision

in order to, for example, adjust the number of spectral envelopes 104 and to turn on / off an optional bit rate detector and adjust the crossover frequency fx. The audio signal 105 is input to the comulation decision unit 220 and is entered into the stereo coding 101 so that the stereo coding 101 can produce the sample values entered in the bandwidth extension unit 230.

Depending on the decision 108 ’generated per switching unit decision unit 220, the tool

The bandwidth extension 230 will generate the spectral band replication data which, in turn, is sent to the audio encoder 210b or the voice encoder 210a.

[0065] The switching decision signal 108 ’depends on a signal and can be obtained from the unit of

switching decision 220 analyzing the audio signal, for example, using a bit rate detector or other detectors that may or may not comprise a variable threshold value. Alternatively, the switching decision signal 108 ’can be adjusted manually (for example by a user) or obtained from a sequence of data (included in the audio signal).

[0066] The output of the audio encoder 210b and the voice encoder 210a can again be entered into the formatter of bit sequence 800 (see Fig. 5).

[0067] Fig. 9b shows an example for the switching decision signal 108 ’which detects a signal of

audio for a period of time before a first time ta after a second time tb. Between the first time ta and the second time tb, the switching decision unit 220 detects a voice signal that gives

as a result different discrete values for the switching decision signal 108 ’.

[0068] The decision using a higher fx crossover frequency is controlled by the switching decision unit 220. This means that the described procedure is further usable within a system where the SBR module is combined with a single core encoder. and crossover frequency fx variable.

[0069] Although some of Figs. 1 to 9 are illustrated as block diagrams of an apparatus, these figures simultaneously are an illustration of a procedure, where the functionalities of the block correspond to the steps of the procedure.

[0070] Fig. 10 illustrates a representation for an encoded audio signal 102 comprising a first portion 104a, the second portion 104b, a third portion 104c and a fourth portion 104d. In this representation the encoded audio signal 102 is a bit sequence transmitted by a transmission channel which further comprises the information of the encoding mode 108. Each portion 104 of the encoded audio signal 102 may represent a different portion of time, although different portions 104 may be in the frequency domain as in the time domain so that the encoded audio signal 102 cannot represent a timeline.

[0071] In this example the encoded audio signal 102 further comprises a first information of the coding mode 108a identifying the coding algorithm used for the first portion 104a; a second information of the coding mode 108b identifying the coding algorithm used for the second portion 104b; a third information of the coding mode 108d identifying the coding algorithm used for the fourth portion 104d. The first information of the coding mode 108a can also identify the first crossover frequency fx1 used within the first portion 104a, and the second information of the coding mode 108b can also identify the second crossover frequency fx2 used within the second portion 104b . For example, within the first portion 104a the mode of

"voice" coding can be used within the second portion 104b the "music" coding mode

It can be used so that the first crossover frequency fx1 can be greater than the second crossover frequency fx2.

[0072] In this example the encoded audio signal 102 does not comprise encoding mode information for the third portion 104c indicating that there is no change in the encoder and / or crossover frequency fx used between the first and third portions 104a, C. Therefore, the information of the coding mode 108 may appear as a header only for the portions 104 that use a different core encoder and / or crossover frequency compared to the preceding portion. In other examples instead of signaling the values of the crossover frequencies for the different portions 104, the information of the encoding mode 108 may comprise a single bit indicating the core encoder (first or second encoder 210a, b) used for the respective portion 104.

[0073] Therefore, the signaling of the switch behavior between the different SBR tools can be performed by presenting, for example, as a specific bit within the bit sequence, so that this specific bit can turn on or off a specific behavior in the decoder. . Alternatively, in systems with two core encoders according to embodiments, switch signaling can also be initiated by analyzing the core encoder. In this case, the presentation of the adaptation of the SBR tools is made implicitly, which means that it is determined by the activity of the corresponding core encoder.

[0074] More details of the standard description of the bit sequence elements for the SBR payload can be found in ISO / IEC 14496-3, subclause 4.5.2.8. A modification of this standard bit sequence comprises an extension of the index to the master frequency table (to identify the crossover frequency used). The index used is coded, for example, with four bits that allow the variable crossover band over a range of 0 to 15 bands.

[0075] The embodiments of the present invention can then be synthesized as follows. Different signals with different time / frequency characteristics have different demands on the characteristic in bandwidth extension. Bit rate signals (for example within the voice signal) need a fine temporal resolution of BWE and the crossover frequency fx (the upper frequency edge of the core encoder) should be as high as possible (for example 4 kHz or 5 kHz or 6 kHz). Especially in vocalized speeches, a distorted temporal structure can decrease the perceived quality. Tonal signals require stable reproduction of the spectral components and a combined harmonic pattern of the reproduced high frequency portions. Stable reproduction of the tonal parts limits the bandwidth of the core encoder but does not require a BWE with fine temporal resolution but rather finer spectral resolution. In a switched voice / audio core encoder design, it is possible to use the decision of the core encoder also to adapt the temporal and spectral characteristics of BWE as well as adapt the BWE start frequency (crossover frequency) to the characteristics of the signal. Consequently, the embodiments provide an extension of bandwidth where the decision of the core encoder acts as a criterion of adaptation to the characteristics of bandwidth extension.

[0076] The signaling of the changed BWE start frequency (crossover) can be done explicitly by sending additional information (such as, for example, encoding mode information 108) in the bit sequence implicitly by deriving the crossover frequency fx directly from the core encoder used (in case the core encoder is, for example, signaled within the bit sequence). For example, a lower BWE fx frequency for the transformation encoder (for example audio / music encoder) and a higher one for a time domain (voice) encoder. In this case, the crossover frequency may be in the range of 0 Hz to Nyquist frequency.

[0077] Although some aspects have been described in the context of an apparatus, it is clear that these aspects also represent a description of the corresponding procedure, where a block or device corresponds to a process step or feature of a procedure step. Similarly, the aspects described in the context of a procedure step also represent a description of a corresponding block or item or feature of a corresponding apparatus.

[0078] The encoded audio signal may be stored on a digital storage medium or transmitted on a transmission medium such as a wireless or cable transmission medium such as the Internet.

[0079] Depending on certain implementation requirements, the embodiments of the invention may be implemented in hardware or software. The implementation can be done using a digital storage medium for example a floppy disk, DVD, CD, ROM, PROM, EPROM, EEPROM or FLASH, with control signals capable of being read electronically stored in them, which cooperate (or are able to cooperate) with a programmable computing system such that the respective procedure is carried out.

[0080] Some embodiments according to the invention comprise a data carrier with control signals capable of being read electronically, capable of cooperating with a programmable computing system such that one of the procedures described is carried out.

[0081] Generally, the embodiments of the present invention can be implemented as a computer program product with a program code, the program code is operative to develop one of the procedures when the computer program product is used in a computer. The program code can for example be stored in a carrier capable of being read by a machine.

[0082] Other embodiments comprise the computer program to develop one of the procedures described herein, stored in a carrier capable of being read by a machine.

[0083] In other words, one embodiment of the invention procedure consists, therefore, of a computer program with a program code to develop one of the procedures described herein, when the computer program is used in a computer.

[0084] Another embodiment of the method of the invention consists, therefore, of a data carrier (or digital storage medium, or medium capable of being read by a computer) comprising, the computer program recorded therein, for develop one of the procedures described herein.

[0085] Another embodiment of the method of invention therefore consists of a sequence of

5 data or sequence of signals representing the computer program to develop one of the procedures described herein. The data sequence or signal sequence may for example be configured to be transferred through a data communication connection, for example over the Internet.

[0086] Another embodiment comprises a processing means, for example a computer, or a programmable logic device, configured or adapted to develop one of the procedures described in

10 this.

[0087] Another embodiment comprises a computer with a computer program installed therein, to develop one of the procedures described herein.

[0088] In some embodiments, a programmable logic device (for example an array of programmable doors per field) can be used to develop one of the functionalities of the procedures

15 described herein. In some embodiments, an array of field-programmable doors can cooperate with a microprocessor in order to develop one of the procedures described herein. Generally, the procedures are preferably developed by any hardware apparatus.

[0089] The embodiments described above are merely illustrative for the principles of

20 present invention. It is understood that the modifications and variations of the arrangements and details described herein will be apparent to those skilled in the art. Therefore, it is intended to limit only by the scope of the impending patent claims and not by the specific details presented by way of description and explanation of the embodiments of the present

Claims (7)

one.

 An apparatus for decoding (100) an encoded audio signal (102), the encoded audio signal (102) comprises a first portion (104a) encoded according to a first coding algorithm, a second portion (104b) encoded according with a second coding algorithm, BWE bandwidth extension parameters (106) for the first portion (104a) and the second portion (104b) and an encoding mode information (108) indicating a first decoding algorithm or a second decoding algorithm, comprising:

A first decoder (110a) stops decoding the first portion (104a) according to the first decoding algorithm for a first time portion of the encoded signal (102) to obtain a first decoded signal (114a), in which the first decoder (110a) comprises an LPC based encoder; A second decoder (110b) for decoding the second portion (104b) according to the second decoding algorithm for a second time portion of the encoded signal (102) to obtain a second decoded signal (114b), in which the second decoder (110b) comprises a transform-based encoder; A controllable BWE module (130) with crossover frequency (fx), the BWE module (130) configured to develop a bandwidth extension algorithm using the first decoded signal (114a) and the BWE parameters (106) for the first portion (104a), and to develop a bandwidth extension algorithm using the second decoded signal (114b) and the bandwidth extension parameter (106) for the second portion (104b), in which the BWE module (130) is configured to use a crossover frequency for bandwidth extension for the first encoded signal (114a) and to use a second crossover frequency for bandwidth extension for the second decoded signal (114b), in which the first crossover frequency is greater than the second crossover frequency; and a controller (140) for controlling the crossover frequency (fx) for the BWE module (130) in accordance with the coding mode information (108).

2.

 The decoder apparatus (100) of claim 1, further comprising an input interface (900) for entering the encoded audio signal (102) as a sequence of bits.

3.

 The decoder apparatus (100) of claim 1 or claim 2, wherein the BWE module (130) comprises a switch (132) configured to switch between the first and second portion of time from the first decoder (110a) to the second decoder (110b) so that the bandwidth extension algorithm is applied to the first decoded signal (114a) or the second decoded signal (114b).

Four.

 The decoder apparatus (100) of claim 3, wherein the controller (140) configured to control the switch (132) depends on the decoding algorithm indicated within the encoding mode information (108).

5.

 The decoder apparatus (100) of one of the preceding claims, wherein the controller (140) is configured to increase the crossover frequency (fx) within the first portion of time or to decrease the crossover frequency (fx) within the Second portion of time.

6.

 A method for decoding an encoded audio signal (102), the encoded audio signal (102) comprises a first portion (104a) encoded according to a first encoding algorithm, a second portion (104b) encoded according to a second Encoding algorithm, BWE bandwidth extension parameters (106) for the first portion (104a) and the second portion (104b) and an encoding mode information (108) indicating a first decoding algorithm or a second algorithm decoding, including the procedure:

the decoding of the first portion (104a) according to the first decoding algorithm for a first time portion of the encoded signal (102) to obtain a first decoded signal (114a), in which the stage of decoding the first portion comprises employing an encoder based on LPC; decoding of the second portion (104b) according to the second decoding algorithm for a second time portion of the encoded signal (102) to obtain a second decoded signal (114b), in which the stage of decoding the second portion (104b) comprises using a transform based encoder; develop a bandwidth extension algorithm by a BWE module (130) with controllable crossover frequency (fx), using the first decoded signal (114a) and BWE parameters (106) for the first portion (104a), and develop , by the BWE module (130) with controllable crossover frequency (fx), a bandwidth extension algorithm using the second decoded signal (114b) and the bandwidth extension parameter (106) for the second portion ( 104b); wherein a first crossover frequency is used for bandwidth extension for the first encoded signal (114a) and a second crossover frequency for bandwidth extension for the second decoded signal (114b), in which the first crossover frequency is greater than the second crossover frequency; and control the crossover frequency (fx) for the BWE module (130) in accordance with the coding mode information (108). 7. A computer program for developing, when used in a computer, the method of claim 6.